The current proposal is an extension of the project to develop methods for objective characterization, recognition and classification of cells by computer analysis of their digitized images which are extended to the next higher level, namely to the classification of entire cytodiagnostic samples. Sample profiles characterizing the cellular composition, state, degree of atypia, and developmental trends of cell samples taken from patients with various clinical cytologic conditions are being analyzed. The potential for the detection of prognostic clues is being explored. A base line study for a taxonomy of preneoplastic cytologic conditions will be made. Classification rules for sample profiles from patients with conditions of cellular atypia are being developed. Work on the further development of the recognition of individual cells by the TICAS method will be continued. Recursive feature extraction methods, and new cell image representation by feature matrices with probabilistically defined elements will be implemented.